Building Intelligent Search Systems With Agentic RAG and OpenSearch
summary
Demonstrates an Agentic RAG pipeline that uses OpenSearch vector DB with Amazon Titan embeddings, an agent built on Strands agent SDK, and a Lambda/API Gateway backend that stores sessions in DynamoDB. Agent composes and runs queries (agent-wrote hybrid queries combining BM25+cosine scoring), retrieves and augments context, and executes tool calls; demo finds surprising real results (a three‑bed under $500k in Lancaster CA), captures session history, and shows the flow is functional and extensible with geo and satellite tools for verification.
AI-written summary · about
chapters (15)
- 0:00Intelligent search systems intro
- 1:24Customer challenges and data
- 3:30Search challenges and OpenSearch
- 5:42Evolution of search tech
- 7:48From intelligent to generative search
- 9:36Use cases beyond chatbots
- 11:45RAG fundamentals
- 13:09Vectorization and embeddings
- 17:02Keyword search foundations
- 19:31Semantic search and vectors
- 20:57Hybrid search techniques
- 23:03Agentic flows with Strands SDK
- 24:50Architecture and data pipeline
- 27:39Live demo: property search
- 34:00Q&A
related talks
22mOrchestrating Multimodal RAG With Agentic Workflows With OpenSearch
34mImplementing an open-source RAG with OpenSearch
27mElevating AI Applications with OpenSearch's Flow Framework and RAG Tool
25mKeynote: OpenSearch in the Agentic Era
43mHarnessing DeepSeek-R1 and OpenSearch for RAG-Powered Intelligent Info...
40mAgentic Retrieval: Building Self-Optimizing Search Systems
27mRio Tinto unlocks potential of generative AI
42mBuilding Self-Healing Infrastructure: Agentic AI & Vector Search in OpenSearch